Northwestern Events Calendar

May
13
2022

Comparative Politics Workshop: Bayesian Integrative Meta Analysis

When: Friday, May 13, 2022
12:00 PM - 1:30 PM Central

Audience: Faculty/Staff - Graduate Students

Contact: Ariel Sowers   (847) 491-7454

Group: Department of Political Science

Category: Academic

Description:

Please join the Comparative Politics Workshop as they host Sarah Moore from Northwestern University. 

Abstract: Across various academic disciplines, meta-analysis is a sophisticated and important tool for the purposes of cumulative knowledge production. While still relatively new to political science analysis, the tool has already contributed interesting discussion to long-standing questions in the field. While debates over the relationship between important political issues such as the relationship between democracy and economic growth or the impact of exposure to political violence have certainly not been solved, meta-analysis has allowed for summative conclusions to be made of these bodies of literature, bridging together studies from disparate contexts. However, the integration of meta-analysis into the modern social sciences is ultimately short-sighted as meta-analysis is so far merely a quantitative endeavor, foregoing systematic inclusion of qualitative scholarship. While some qualitative methods exist to mirror the same types of cumulative assessments that meta-analysis offers quantitative scholarship, there is not yet a method concerned with the systematic assessment of qualitative and quantitative scholarship in tandem. To address this problem, this project proposes a method of Bayesian Integrative Meta-Analysis (BIMA). I propose BIMA as a method that systematically includes qualitative scholarship in meta-analysis through converting qualitative research into Bayesian prior information. I argue that extracting expert information from qualitative, scientific manuscripts that meet inclusion criteria for a meta-analytical study can be treated as a Bayesian elicitation process. This elicited expert information can then be systematically incorporated into a prior distribution. The resulting posterior distribution of the Bayesian meta-analysis is then then result of a qualitatively informed prior and a likelihood distribution composed of the quantitative studies in the meta-analysis. In the full paper, I will show the usefulness of this new methodological tool alongside a test case, developing a novel meta-analysis regarding the effect of conflict exposure on individual-level sociopolitical outcomes. I will show how meta-analysis estimates may differ between the merely quantitative meta-analysis and the integrative meta-analysis. However, this presentation will address and seek comments on the BIMA framework and different coding procedures relevant to the meta-analysis.

Sarah Moore is currently a PhD candidate in Comparative Politics and Master of Science student in Statistics at Northwestern University. Her broad interests include the legacies of political violence, mixed-methods research, and civilian resistance in Latin America.

Moore's dissertation is a collection of essays that devise quantitative and qualitative methods for the improved study of difficult-to-research phenomena, with specific applications to the study of organized violence. Informed by her additional training in psychometric theory and sociological qualitative methods, these four essays provide methodological solutions for sampling, measurement of concepts, and inference of preferences where randomization is not possible. Her research is supported by the Buffett Institute for Global Affairs at Northwestern University. She has also served as a Predoctoral Research Fellow with Universidad de los Andes in Bogotá.

 

Register Add to Calendar

Add Event To My Group:

Please sign-in